K-means Evaluation - control comparisons and manual 1st 1K correlations
JL
2/16/2018
K-Means notes
- K-means is currently trying to identify 4 clusters using 3 dimensions (CCR7, CD45R4, and CD28)
- The input to k-means is currently the union of Helper and Cytotoxic T cells
- For current results below, only OpenCyto was used to determine input for kmeans clustering (i.e did not start with Helper/Cyto T defined by manual .wsps)
Control gating “consistency”
- OC controls were filtered for likely failures (lose ~%10 of the samples with this filter)
Control summary for results
| CONTROL | N_FCS |
|---|---|
| A | 10 |
| B | 6 |
| C | 9 |
| D | 7 |
| E | 31 |
| F | 18 |
| G | 20 |
| H | 24 |
| KEYFCS | SAMPLE | POPULATION | COUNT | PARENT_COUNT | OPTIMAL_K | DFK_ALL | NUM_POPS_ASSIGNED | TOTAL_COUNTS | QC | PANEL | MACHINE | LAB_ID | formatName | DATE | DATE_MONTH | DATE_WEEK | formatExp | EXPERIMENTER | SANITIZE_NAME | MANUAL_ANNOTATION | PARENT_FREQ | PARENT_PERCENT | KEY | CTL | Study.ID | variable | value |
|---|
Correlations to manual 1st 1k
